The long-term objective is to understand the oculomotor system. The applied goals are to provide data for more quantitative and sensitive clinical tests that will help in determining eh locations of brain lesions. We also seek the basic mechanisms by which the nervous system repairs dysmetria after disease or trauma. The basic goals are to discover how the nervous system processes neural signals in regulating the control of movement; in particular, how this is accomplished by networks of neurons that can learn. Four initial projects are aimed at these goals. The first uses a new theoretical approach: a backward-propagating model of a learning neural network. A major feature of this network is that the tasks are distributed among its neurons just as we observed in the oculomotor system. Pursuit, saccadic and vestibular commands are distributed over the vestibulaR-oculomotor regions of the caudal pons. Action directions of second-order anal afferents are distributed over all directions as are the action directions of burst neurons that create saccades. These patterns can be accounted for with this learning network. This is the first attempt to model at the nerve-network level in the oculomotor system and we hope that it will initiate a major move in this direction. A second project addresses the neural integrator which changes eye-velocity into eye-position commands. We have located it in the nucleus prepositus hypoglossi and vestibular nucleus and believe that it depends heavily on commissural connections. We will investigate this in the alert monkey by microstimulation and micro injections of a local anesthesia in this region to provide evidence that may suggest how the neural integrator works. A third project investigates predictive tracking of the saccadic and pursuit systems and how they interact. We first explore how many state changes (change in target position or velocity for given durations) can be stored in their internal pattern generators. We next ask to what extent random fluctuations in one system interfere with the ability to predict but we have very little by way of a normal data base. The fourth project involves motor learning in the control of the gain of the vestibulo-ocular reflex (VOR) in the cat with emphasis on the role of climbing fibers. We intend to record form Purkinje cells in the flocculus and measure the gain of the VOR in normal or gain-adapted animals while either stimulating or anesthetizing the climbing fibers. We hope to provide information to test the hypothesis that motor learning is stored in the cerebellum.